The objective of this study was to analyze the mortality during the first 24 months after delivery in relation to CD4 T-lymphocyte levels and viral load at enrollment (36 weeks of gestation) in a cohort of HIV-1-seropositive breast-feeding women at the Dar es Salaam site of the multicenter Petra trial (a mother-to-child HIV-1 transmission intervention trial using antiretroviral therapy). Antiretroviral treatment was not available in this setting apart from the short treatment given within the trial around delivery to prevent mother-to-child transmission of HIV. T-lymphocyte subsets were determined by flow cytometry. Plasma HIV-1 RNA was quantified by the Amplicor HIV-1 RNA Monitor v 1.5 assay. Mortality after delivery was analyzed using the life-table technique and Cox regression. The analysis included 266 mothers. The CD4 cell counts at enrollment were <200 cells/mm3 in 14.5% of the mothers. The viral load at enrollment was >100,000 RNA copies/mL in 33.6% of the mothers. The mortality 24 months after delivery was 6.7% (95% CI = 3.1-10.1%). The mortality 24 months after delivery was 29.9% (95% CI = 13.1-46.9%) for mothers with <200 CD4 cells/mm3 at enrollment, 3.3% (95% CI = 0-6.6%) for mothers with 200-499 CD4 cells/mm3, 2.9% (95% CI = 0-7.1%) for mothers with >500 CD4 cells/mm3 (P = 0.0000), 15.0% (95% CI = 6.6-23.4%) for mothers with viral load >100,000 copies/mL at enrollment, and 2.8% (95% CI = 0-5.6%) for mothers with viral load <100,000 copies/mL (P = 0.0000). In the multivariate analysis CD4 cell counts and viral load were both independent risk factors for mortality (P < 0.001 and P = 0.004, respectively). In conclusion, the mortality was high among women with severe immunosuppression or high viral load at enrollment, but not in the rest of the women. CD4 lymphocyte count in late pregnancy was a better predictor of death within 2 years than was viral load. The results support the World Health Organization recommendation to initiate antiretroviral treatment in resource-limited settings in HIV-1-infected adults with CD4 cell counts <200/mm3 and show that this is appropriate also among perinatal women.
From the Departments of *Obstetrics and Gynecology, †Internal Medicine, ‡Microbiology and Immunology, Muhimbili University College of Health Sciences, Dar es Salaam, Tanzania; and §Swedish Institute for Infectious Disease Control and ¶Microbiology and Tumor Biology Center, Karolinska Institute, Stockholm, Sweden.
Received for publication March 9, 2004; accepted July 26, 2004.
This work was performed within the framework of the UNAIDS/WHO-coordinated multicenter Petra trial. The Petra trial at the Dar es Salaam site received financial support from the Swedish International Development Agency (Sida), Department of Research Cooperation (SAREC).
Reprints: Gunnel Biberfeld, Swedish Institute for Infectious Disease Control and Karolinska Institute, SE-171, 82 Solna, Sweden (e-mail: email@example.com).
The United Nations Program on HIV/AIDS (UNAIDS) estimates that at the end of 2003 there were 26.6 million people living with HIV infection in sub-Saharan Africa.1 About 60% of the HIV-infected adults are women. AIDS is the number one cause of death in sub-Saharan Africa.
Conflicting data have been presented on the mortality among breast-feeding HIV-1-infected women in sub-Saharan Africa. In a study in Nairobi, Kenya, the mortality 24 months after delivery was higher among HIV-1-infected breast-feeding mothers than among HIV-1-infected mothers who formula fed their infants.2 However, in Durban, South Africa, no deleterious effects of breast-feeding on the health of HIV-1-infected women were observed.3
It is well documented through studies in industrialized countries that CD4 T-lymphocyte counts and plasma viral load are good predictors of progression to AIDS and death.4,5 However, information about the value of these 2 markers for predicting mortality in HIV-1-infected individuals in African countries is limited.2,6,7
A multicenter, randomized, placebo-controlled trial (the Petra trial) that used a combination of zidovudine and lamivu-dine in 3 short-course regimens to determine the efficacy in preventing mother-to-child transmission (MTCT) of HIV-1 was conducted in Uganda, South Africa, and Dar es Salaam, Tanzania.8 In the present study we have analyzed the mortality during the first 24 months after delivery in relation to CD4 T-lymphocyte levels and viral load at enrollment in the cohort of HIV-1-infected mothers enrolled into the Petra trial at the Dar es Salaam site.
SUBJECTS AND METHODS
The study population consisted of women enrolled at the Dar es Salaam site of the UNAIDS coordinated multicenter Petra trial on prevention of MTCT of HIV-1 transmission through antiretroviral treatment; the details are described elsewhere.8 Briefly the women were recruited from 3 antenatal clinics in Dar es Salaam where voluntary counseling and HIV testing were offered to all pregnant women before 34 weeks of gestation. Pretest and posttest counseling were given by trained midwives. The eligibility criteria included HIV-1 sero-positivity confirmed on a second test, hemoglobin level not less than 8 g/dL, being ≥18 years of age, availability for 18 months of follow-up, and being an accessible resident of Dar es Salaam. The women had to give written informed consent to participate in the Petra trial and were free to withdraw from the trial at any stage if they wished.
The women were enrolled at week 36 of gestation. They received normal antenatal, labor, and delivery care, like the rest of the pregnant women. The clinical status of the women at enrollment was determined using the World Health Organization (WHO) clinical staging system for HIV infection and disease.9 The enrolled women received antiretroviral/placebo treatment according to the Petra trial randomization. The trial was double-blind. The antiretroviral/placebo was administered in 4 short regimens.8 In arm A, the study drugs (zidovudine and lamivudine) were administered from 36 weeks' gestation, intrapartum, and 1 week postpartum (to mother and child); in arm B, study drugs were administered intrapartum and postpartum; and in arm C, intrapartum only. Arm D was a placebo arm that was discontinued immediately after February 1998 when publication of results from a short-course, randomized, placebo-controlled trial in Thailand became available.10
All mothers were requested to deliver at Muhimbili hospital. Postnatally mothers and infants received free medical care from obstetricians and pediatricians, including laboratory monitoring tests. A special postnatal clinic was set up near a regular pediatric clinic and follow-up appointments were given at weeks 1, 3, and 6 and at months 3, 6, 9, 12, 15, 18, 21, and 24 months after delivery.
Home tracing of study subjects was conducted if they missed 2 consecutive appointments. Mothers who lost their children continued to be followed, and for those who moved upcountry information was obtained from relatives and friends.
Antiretroviral treatment was not available in this setting apart from the short regimen given within the trial around delivery to prevent MTCT of HIV. Breast milk substitute was not offered to the women.
HIV antibodies were determined by 2 consecutive anti-HIV enzyme-linked immunosorbent assays (ELISAs), Enzygnost anti-HIV 1 + 2 Plus Elisa (Behring, Marburg, Germany) and Wellcozyme HIV-1 recombinant ELISA (Murex, Dartford, UK). Sera reactive on both tests were considered HIV-1 antibody positive and those with repeatedly discordant results on ELISA were tested by a Western blot assay. A second sample was collected for confirmation of initially reactive tests before the final results were made available to the women.
Determination of T-lymphocyte subsets was done using the SimulSET flow cytometry method (Becton Dickinson, Immunocytometry System, San Jose, CA) as described previously.11
Plasma HIV-1 RNA was quantified by the Amplicor HIV-1 RNA Monitor v1.5 assay (donated by Roche Diagnostic Systems).
Data analysis was done using the SPSS software system 11.0 (Statistical Package for Social Sciences, SPSS, Inc., Chi-cago, IL). Differences between groups of observations with skewed distributions were tested by the Kruskal-Wallis test. Mortality and breast-feeding were analyzed by the life-table technique, taking censored observations into account. Women lost to follow-up were censored at the time for their last visit to the clinic. Wilcoxon-Gehan statistic was used testing differences between groups. Univariate analyses were also carried out with Cox regression. The forward stepwise Cox regression model was used in the multivariate analyses.
The main study protocol was approved by the ethics committee of the Tanzania National AIDS Control Program and the Research and Publication Committee of Muhimbili University College of Health Sciences, Dar es Salaam, and the Institutional Review Board of the Karolinska Institute, Stock-holm, Sweden.
Of 288 HIV-1-infected pregnant women enrolled into the Petra trial at the Dar es Salaam site, 2 women died, 2 disappeared, and 3 refused to continue before delivery. One woman died during delivery. For 14 women there was no information available after delivery. The remaining 266 mothers were included in the analysis of mortality 24 months after delivery. During the follow-up 15 mothers died, 65 were lost to follow-up before month 24, and 186 were still in follow-up 24 months after delivery.
Among the 266 mothers in the mortality analysis, the median age was 25 years. At enrollment 83.1% were classified as having stage 1 disease according to the WHO staging list, 8.3% as stage 2, 7.5% as stage 3, and 1.1% as stage 4. The CD4 T-lymphocyte absolute counts at enrollment were <200 cells/mm3 in 14.5% of the women, 200-499 cells/mm3 in 53.2%, and >500 cells/mm3 in 32.3%. The median CD4 T-lymphocyte absolute count at enrollment was 410 cells/mm3 and the median CD4 T-lymphocyte percentage 25%. Plasma viral RNA levels ≥100,000 copies/mL were demonstrated in 33.6% of the women. The median plasma HIV-1 RNA level at enrollment was 42,658. The median CD4 T-lymphocyte counts and viral load according to WHO clinical stage are shown in Table 1. There was a significant decrease of CD4 T-cell counts as well as percentages between stage 1 and stage 2 and between stage 1 and stage 2, 3, 4 grouped together, but there was no meaningful trend when all stages were compared individually. The viral load values also differed significantly between WHO stage 1 and stages 2-4.
At week 1 after delivery, 96% of the mothers breast-fed their infants and at 24 months 7.6% were still breast-feeding. The median length of breast-feeding was 11.2 months. The women were considered to have stopped breast-feeding at the time of complete weaning. No information was available regarding to which extent exclusive breast-feeding was practiced.
The vital status at 24 months after delivery was known for 201 of the 266 mothers included in the mortality analysis. The mean length of follow-up was 19.0 months (SD = 8.5). Of the 65 mothers with unknown vital status 24 months after delivery, 46 mothers had moved (mainly upcountry to their home villages), 7 mothers did not want to continue follow-up, and the rest could not be traced. Mothers lost to follow-up before month 24 did not differ significantly from the rest of the mothers regarding WHO stage, CD4 T-lymphocyte counts, or viral load at enrollment. Regarding age there was a tendency that mothers lost to follow-up before month 24 were younger than the rest of the women (data not shown).
The overall mortality 24 months after delivery was 35.6/1000 person-years. In the survival analysis, the cumulative mortality 24 months after delivery was 6.7% (95% CI = 3.1-10.1%). The mortality 24 months after delivery was 29.9% (95% CI = 13.1-46.9%) for mothers with <200 CD4 cells/mm3 at enrollment, 3.3% (95% CI = 0-6.6%) for mothers with 200-499 CD4 cells/mm3, 2.9% (95% CI = 0-7.1%) for mothers with >500 CD4 cells/mm3 (P < 0.001) (Fig. 1A), 15.0% (95% CI = 6.6-23.4%) for mothers with viral load >100,000 copies/mL at enrollment, and 2.8% (95% CI = 0-5.6%) for mothers with viral load <100,000 copies/mL (P < 0.001) (Fig. 1B). In the univariate analysis there were also significant differences in mortality according to hemoglobin value and WHO stage at enrollment. Independent risk factors for mortality in the multivariate analysis were CD4 lymphocyte count (P < 0.001) and viral load (P = 0.004) (Table 2). If viral load was excluded from the regression analysis, hemoglobin was an independent risk factor together with CD4 cell counts (data not shown).
The short antiretroviral treatment given to mothers in our cohort to prevent MTCT of HIV-1 within the Petra trial was not expected to have any effect on the mothers' health. The length of treatment varied from a few weeks to no treatment in 4 different treatment arms. There was no statistically significant difference in mortality according to treatment arms (data not shown).
Of the 15 deaths registered in our cohort during the first 24 months after delivery, 12 women (80%) met the clinical criteria for AIDS (WHO stage 4) at death, while at enrollment only 1 woman was classified as being in WHO stage 4, 2 in WHO stage 3, 2 in WHO stage 2, and the rest in WHO stage 1. However, if CD4 cell levels are taken into account, 9 of the 15 mothers who died fulfilled the laboratory or clinical criteria for AIDS at enrollment. Tuberculosis appeared among causes of death for 5 of the 15 women who died. Two mothers who died had developed Kaposi sarcoma.
The overall mortality in our cohort of HIV-1-infected women in Dar es Salaam compares well to other studies on mortality in HIV-1-infected women in Africa.12,13 The 24-month mortality in our cohort lies between the 24-month mortality rates reported from Nairobi by Nduati et al2 for breast-feeding and for nonbreast-feeding HIV-1-infected women (11 and 4%, respectively) and is not significantly different from either of them. The median length of breast-feeding in our study was 11.2 months, compared with a median time of breast-feeding of 17 months in the breast-feeding group in the Nairobi study. The median plasma viral load and the median CD4 cell count at enrollment were similar in the women in our cohort and in the women in the Nairobi study.
As the viral load levels were generally high in our cohort, we chose to use 100,000 copies/mL as the dividing line for our grouping. Still, as many as 33.6% of the women had viral load values above that level and even with this high dividing line for the grouping, viral load did not predict mortality within 2 years as well as the CD4 cell level did. It has been reported that with increasing duration of infection the predictive value of low CD4 T-cell counts for progression to AIDS increases whereas the predictive value of high plasma HIV-1 RNA levels decreases.5 The duration of infection among the women in our study was not known.
In the univariate analyses we found a significant difference in mortality according to WHO stage and according to hemoglobin level, but in the multivariate analysis only the CD4 cell level and the viral load remained as independent predictors of mortality. However, as severe anemia was an exclusion criterion at enrollment, the possibility of evaluating hemoglobin as a predictor of mortality in our cohort was limited.
Our data show that most of the mothers who died within 24 months after delivery were women who were already severely immunosuppressed or who had a very high viral load when recruited into the study in late pregnancy. The study of HIV-1-infected women in Nairobi also showed a significant association between mortality and CD4 cell counts and viral load at enrollment before delivery.2
In our cohort most of the mothers who died (80%) had clinical AIDS at death. This differs from findings in a study from Rwanda, in which only half of the HIV-1-infected women who died fulfilled the clinical criteria for AIDS.12 Most of the mothers who died in our study had a rapid deterioration of their clinical status, from being clinically well at enrollment to developing clinical AIDS and then dying within a few months. This is in agreement with earlier studies showing that HIV-1-infected people in Africa progress fast from AIDS to death, as did HIV-1-infected people in industrialized countries before the use of antiretroviral therapy.14 Tuberculosis was a major cause of mortality in our cohort, which is also in agreement with findings in other African settings.6,13
In conclusion, we have found that the CD4 lymphocyte count in late pregnancy was a better predictor of death within 2 years after delivery than was viral load in this cohort of HIV-1-infected women in Dar es Salaam. Viral load, however, showed independent impact on survival, taking CD4 cell levels into account. Clinical status in late pregnancy alone was not a good predictor of death within 2 years. The mortality was high among women who had severe immunosuppression at enrollment in the study (30% in those with <200 CD4 cells/mm3) or who had a very high viral load, but not in the others, despite the fact that they breast-fed their babies. The mortality was similar (3%) in women with 200-499 CD4 cells/mm3 and in women with >500 CD4 cells/mm3. Our results support the WHO recommendations15 to start antiretroviral treatment in resource-limited settings in HIV-1-infected adults with CD4 cell counts <200/mm3 and show that this is appropriate also among perinatal women. Our findings also support the WHO standpoint that, in resource-limited settings, measurement of viral load is not necessary before starting antiretroviral therapy. The majority of the immunosuppressed women in our study would have benefited from access to antiretroviral treatment to improve their health. Apart from being given access to programs to prevent MTCT of HIV, HIV-infected pregnant women should have access to care and support programs, which include prevention and treatment of opportunistic infections, psychosocial and nutritional support, and treatment with antiretroviral drugs for immunosuppressed women.
The authors thank Ephraim Mbena, Eva Olausson-Hansson, and Anita Östborn for excellent technical assistance and Bo Nilsson for statistical assistance. We also thank the nurses/midwives and all other personnel involved in the study.
1. UNAIDS/WHO 2003. AIDS Epidemic Update
2003. Geneva. UNAIDS.
2. Nduati R, Richardson BA, John G, et al. Effect of breastfeeding on mortality among HIV-1 infected women. Lancet
3. Coutsoudis A, Coovadia H, Pillay K, et al. Are HIV-infected women who breastfeed at increased risk of mortality? AIDS
4. Mellors JW, Muñoz A, Giorgi JV, et al. Plasma viral load and CD4+
lymphocytes as prognostic markers of HIV-1 infection. Ann Intern Med
5. Wolf Fd, Spijkerman I, Schellekens PTh, et al. AIDS prognosis based on HIV-1 RNA, CD4+
T-cell count and function: markers with reciprocal predictive value over time after seroconversion. AIDS
6. French N, Mujugira A, Nakiyingi J, et al. Immunologic and clinical stages in HIV-1-infected Ugandan adults are comparable and provide no evidence of rapid progression but poor survival with advanced disease. J Acquir Immune Defic Syndr
7. Morgan D, Rutebemberwa A, Malamba S, et al. HIV-1 RNA levels in an African population-based cohort and their relation to CD4 lymphocyte counts and World Health Organization clinical staging. J Acquir Immune Defic Syndr
8. The Petra team. Efficacy of three short-course regimen of Zidovudine and Lamivudine in preventing early and late transmission of HIV-1 from mother to child in Tanzania, South Africa and Uganda (Petra study): a randomized, double-blind, placebo-controlled trial. Lancet
9. World Health Organization. Acquired immune deficiency syndrome (AIDS): interim proposal for a WHO staging system for HIV infection and disease. Wkly Epidemiol Rec
10. Shafer N, Chuachoowong R, Mock PA, et al. Short course zidovudine for perinatal HIV transmission in Bangkok, Thailand: a randomised placebo-controlled trial. Lancet
11. Urassa WK, Mbena EM, Swai AB, et al. Lymphocyte subset enumeration in HIV seronegative and HIV-1 seropositive adults in Dar es Salaam, Tanzania: determination of reference values in males and females and comparison of two flow cytometric methods. J Immunol Methods
12. Lindan CP, Allen S, Serufilira A, et al. Predictors of mortality among HIV-infected women in Kigali, Rwanda. Ann Intern Med
13. Leroy V, Msellati P, Lepage P, et al. Four years of natural history of HIV-1 infection in African women: a prospective cohort study in Kigali (Rwanda), 1988-1993. J Acquir Immune Defic Syndr Hum Retrovirol
14. Morgan D, Mahe C, Mayanja B, et al. HIV-1 infection in rural Africa: is there a difference in median time to AIDS and survival compared with that in industrialized countries? AIDS
15. World Health Organization. Scaling Up Antiretroviral Therapy in Resource-Limited Settings: Treatment Guidelines for a Public Health Approach: 2003 Revision
. World Health Organization: 2004.
Keywords:© 2005 Lippincott Williams & Wilkins, Inc.
HIV; women; mortality; survival; Africa